Spontaneous recovery after stroke is generally incomplete, making post-stroke disability a highly prevalent problem. New therapies are in development to reduce disability in patients in the chronic phase of stroke. Such therapies aim to improve behavioral status not by acutely salvaging tissue but by changing function in brain elements that survive to the chronic phase. This restorative approach has the potential to reach a high fraction of subjects with stroke. Limited guidance is available for optimal prescription of such therapies. Among patients in the chronic plateau phase, which characteristics suggest potential for reducing disability with therapy, and which suggest that deficits are instead fixed? To address this issue, 108 patients who have had a stroke in the last 20 weeks and whose motor status has reached a plateau will be evaluated, in multiple domains, including demographic, clinical, radiological, brain mapping, neurophysiological, and genetic measures. A 3-week course of standardized arm motor therapy will follow, after which clinical assessments related to arm motor function will be rechecked. This study hypothesizes that the ability to predict treatment response using behavioral exam (the usual means of selecting patients in this context) is substantially improved with addition of several measures, particularly of brain injury and function. Analyses will also define an injury threshold that indicates which patients are, or are not, likely to improve with therapy. Other analyses will focus on a proposed model of learned disuse: the ability to identify those patients who are underusing brain regions that survived the stroke would aid chronic stroke therapies. One measure of particular interest is BDNF genotype. The BDNF val66met polymorphism is present in approximately 1/4th of people, has been associated with reduction in certain measures of cortical plasticity, and might be of particular importance to predicting both spontaneous and therapy-induced stroke recovery. This will be evaluated in the 108 patient therapy study, as well as a study of spontaneous recovery in 150 patients across 9 sites. Disability affects stroke patients for many years. A number of new therapies focused on CNS repair might improve this situation. However, optimal approach to applying these therapies is unclear, particularly the issue of selecting those patients most likely to respond to therapy. The overall hypothesis of the proposed study is that characterizing CNS injury and function will provide neurobiological insights useful for optimally applying restorative therapies to human patients with stroke. Such knowledge should facilitate reducing disability in appropriately targeted patients with stroke.